Source notes
How the numbers are built
The site is a live model, not direct observation. Every counter uses the same reusable engine, combining a global population or event baseline with geographic distribution, local time, participation, typical duration and an activity-specific hourly curve.
Core model
baseline estimate = Σ (region population × activity probability at the region’s current local time)
Twelve practical population-weighted UTC bands approximate where humanity is located. A region at 03:00 contributes much more to sleeping than one at 15:00; work, meals, commuting, exercise and evening leisure each use a different curve.
Reference baselines
- World and country population.
United Nations, World Population Prospects 2024 - United States 2024 population benchmark.
U.S. Census Bureau, 2024 population estimates - Air travel: 4.96 billion annual passenger journeys.
International Air Transport Association, 2024 outlook - Births and population structure.
UN World Population Prospects and UNICEF statistical tables - Physical activity context.
World Health Organization physical activity reporting - Internet connectivity context.
International Telecommunication Union Facts and Figures
Editorial assumptions
There is no single worldwide, second-by-second database for most human activities. Eligibility, participation, frequency and duration are rounded editorial assumptions. Activity pages keep source links and model assumptions separate; where source review is incomplete, the page says so instead of inventing a citation.
One person can appear in several estimates at once. Someone may be working, online, listening to music and drinking coffee, so activity totals are not mutually exclusive and must not be added together.
Evidence-gated live context
The site can check current external data, but a verified event and a verified numerical effect are treated as different claims. The current adapters use the official U.S. National Weather Service alerts API and attempt direct player observations documented by the Steamworks Web API. If a live source cannot be reached from the static site, no event is created.
current estimate = baseline estimate + verified numerical adjustment
The adjustment remains zero unless a live event reaches Level 3 and contains an explicit, auditable integration rule. A measured subset that is already represented in the baseline—such as selected Steam players—is not added because that would double-count people.
Evidence levels
- Level 0 — Mention only.
The event is verified, but no activity claim is made. - Level 1 — Live context.
The event is relevant; the estimate is not changed. - Level 2 — Direction supported.
Reliable signals indicate higher or lower activity, but no percentage is invented. - Level 3 — Quantifiable.
A direct numerical observation exists. It changes the estimate only if the integration avoids double counting and has a documented calculation.
Each live event records source publisher and URL, publication/event/update/retrieval times, active period, decay period, expiration, geography, relationship, evidence level and confidence. Every numerical adjustment also writes an internal audit record containing the baseline, raw value, calculation, result and timestamp.
AI is never a data source. It may help classify or summarize sourced events, but it cannot invent an impact magnitude or be cited as evidence.
Country mode
Country rankings use rounded 2024 population anchors, one representative UTC offset and the same global activity assumptions. They are labelled as country models rather than country measurements. The engine is structured so multi-zone and country-specific assumptions can be added later.
Why the values move
The model recalculates from the actual clock. As local time advances across every population band, the hourly probability curve changes continuously. No random number generator is used to create counter movement, rankings or trends.
Trends and the 24-hour chart
“Rising Right Now” compares the current model output with the same model 30 minutes earlier. The 24-hour chart is a typical modelled curve—not a claim about collected historical observations.
Limitations
Broad UTC bands cannot capture every country’s culture, schedule, season, age structure or access. Uncertainty ranges on activity pages communicate that the central number is useful for perspective, not precision.